International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 13

This material is protected by copyright by IGI Global and is intended for your personal use.
Material may not be: used for commercial purposes, sold, included in any open access or
other databases, or be modified and republished without the written consent of IGI Global.

the Bi-Based organization

Barbara Wixom, University of Virginia, USA
Hugh Watson, University of Georgia, USA

aBstraCt
Business intelligence (BI) is an umbrella term that is commonly used to describe the technologies, applica-
tions, and processes for gathering, storing, accessing, and analyzing data to help users make better decisions.
For BI-based irms, BI is a prerequisite for competing in the marketplace. Though there are several possible
BI targets, it is important to understand how they differ in terms of strategic vision, level of sponsorship, re-
quired resources, impact on people and processes, and beneits. Some companies like Harrah’s Entertainment,
Continental Airlines, Norfolk Southern, and Blue Cross and Blue Shield of North Carolina are exemplars of BI
best practices. Despite the progress made with BI, there are still many opportunities for academic research.

Keywords: Analytics, Beneits, Business Intelligence, Case Study, Data Warehousing, Trends

introduCtion to help managers better plan and optimize

specific business goals and activities, such as
The inaugural issue of the International Journal production planning and investment portfolio
of Business Intelligence Research provides an optimization (Power, 2007). The term decision
excellent opportunity to reflect on the history support systems, or DSS, was used to describe
and future of BI, as well as the remarkable and these analytic applications, and later became the
unprecedented transformations it can effect. BI name associated with an academic field.
has a rich history with origins that date back Over the years, a variety of additional deci-
fifty years ago. As we journey ahead in BI, we sion support applications emerged, including
must acknowledge from where we have come executive information systems, expert systems,
and what we currently understand. Only then and online analytical processing, each with spe-
can we further the field in effective ways. cific differentiating characteristics and unique
Organizations first used computers in the names. In the early 1990s, Howard Dresner, a
1960s for transaction processing and scientific Gartner analyst, coined the expression business
applications. At the time, organizations did not intelligence to describe these applications;
focus on decision support, although they did since then the term has come to be widely used
create reports that summarized the transaction in industry, and to a slightly lesser extent in
data that was processed. By the late 1960s, the academia, as an umbrella term for all decision
first decision support applications emerged support applications (Gartner, 2009b). In recent
years, analytics also has been used to describe
DOI: 10.4018/jbir.2010071702

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
14 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

applications that provide decision support a Bi definition

(Davenport, 2006).
Just as the terminology for BI has evolved, There is no universally-accepted definition
so has BI’s role and impact on organizations. of BI, but the following is useful for our pur-
BI has moved from being a contributor to or- poses:
ganizational success to being a prerequisite for
it; indeed, for many firms, BI is a prerequisite Business intelligence (BI) is a broad category
for even competing in the marketplace. We call of technologies, applications, and processes for
these firms BI-based organizations because gathering, storing, accessing, and analyzing
of the critical role that BI plays. BI also has data to help its users make better decisions.
changed from being a tool used by a few special-
ists to one that is used by many workers. It has Several aspects of this definition warrant
changed from focusing solely on the analysis of discussion. While BI sometimes is thought of
historical data to including the capture and use in terms of applications, such as dashboards/
of real-time data to impact current, operational scorecards or predictive analytics, we consider
decisions. Consequently, it is not surprising that BI to be broader in scope. For us, it includes
Gartner found in both 2008 and 2009 that BI gathering data from source systems, storing
was at the top of many CIOs’ strategic agendas the data, and accessing and analyzing the data
(Gartner 2008, 2009a). using BI technologies and applications. Putting
For several decades we have been study- it differently, BI includes both getting data in
ing what leading companies have been doing (to a data mart or warehouse1) and getting data
with BI. We have identified best practices by out (through technologies or applications that
judging The Data Warehousing Institute’s meet some kind of business purpose).
Best Practices Awards competitions and by Much like Sprague’s (1980) distinction
developing case studies about BI and data between DSS applications and DSS genera-
warehousing vendors’ top customers (e.g., tors, BI includes both BI applications and BI
Anderson-Lehman, Watson, Wixom, & Hoffer, technologies, which can be used to develop the
2004; Cooper, Watson, Wixom, & Goodhue, applications. For example, a company may have
2000). Here we share findings which are not dashboards for call center operators, but these
only interesting and significant, but which dashboards are built using a specific technology,
demonstrate the transformative - and sometimes such as MicroStrategy Report Services. Simi-
unanticipated - power of BI. In particular, we larly, a financial analyst may have a financial
describe practices at Harrah’s Entertainment, planning application built using Microsoft Excel
Continental Airlines, Norfolk Southern, and that accesses warehouse data.
Blue Cross and Blue Shield of North Carolina. Processes are also a very important part of
Prior to discussing each company, we provide a BI. For example, there have to be processes for
conceptual foundation for interpreting the work extracting, loading, and storing data; maintain-
that was done. Later we discuss a BI maturity ing metadata for IT and users; and prioritizing
model that helps explicate how organizations’ BI projects. Some of these processes are the
BI capabilities evolve over time. In closing, responsibility of the BI staff, while others are
we provide suggestions for future academic BI the joint responsibility of the BI staff and the
research. We begin, however, with a definition business units.
of BI and a description of a generic best practice A variety of stakeholders play essential BI
BI environment. roles. Extraction, Transformation and Loading
(ETL) experts, data modelers and database
administrators focus on preparing the data
warehouse for use. Over time, data stewards,

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 15

project champions, and business owners drive than just transactional data, such as Web 2.0
governance and quality processes to ensure that collaborations, emails, Word documents, and
BI remains aligned with the business’ strategies third party data (e.g., U.S. postal addresses,
and objectives. Data miners, analysts and busi- customer demographics). Companies extract
ness users extract value from the data. the data from the sources, transform the data
(by matching, integrating, and aggregating
a Best Practice Bi environment it), and load the data into the data warehouse.
This process often is referred to as ETL, but
Figure 1 shows a generic, comprehensive BI with the emergence of additional ways of get-
environment. Not all organizations fully follow ting data into the warehouse, such as real-time
this model. Some companies, for example, may data trickle feeding, the term data integration
have an alternative architecture for their data is increasingly used.
warehouse, such as having no dependent data There are a variety of architectures for
marts. Others may have a less comprehensive storing data in a warehouse, but we present the
environment with no data warehouse, but, hub-and-spoke approach (i.e., a data warehouse
instead, data marts that support specific applica- hub and dependent data marts spokes) because
tions and business units. All of these variations it is the most common (Ariyachandra & Watson,
represent BI, but Figure 1 shows a model that 2005), works well, and is used in many of the
we think exemplifies best practice, and which companies that we later describe.
is typically used by BI-based organizations. At the right of the diagram, we see that a
Starting at the left of the diagram, we see variety of users and applications access ware-
that data originates from a variety of source house data. There are BI information producers
systems, including transactional and ERP (e.g., financial analysts) who analyze data and
systems. Increasingly, sources contain more

Figure 1. A best practice BI environment

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
16 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

create BI information for others, and informa- warehouse, implementing a variety of BI tools
tion consumers (e.g., managers) who use BI and applications, and investing in BI user train-
information created by others (Eckerson, 2002). ing. A third category of companies is trying to
A focal point in many organizations is to make transform itself organizationally, and aims to
the use of BI more pervasive. One way of doing use BI as an enabler for the firm’s new business
this is to make BI available to more people, model. These companies undergo significant
including operational personnel, customers, process change and use BI to run the company
and suppliers. Potential BI applications include differently. Table 1 highlights the differences
SQL queries, drillable reports, OLAP, EIS, among these targets.
dashboards/scorecards, alerts, and data mining/ None of these targets is inherently better
predictive analytics. than any of the others; the selection of the target
It is important to recognize that the BI en- depends on the business needs and BI readi-
vironment includes metadata, data quality, and ness of the organization (Williams, 2004). For
governance. While all three involve technology example, to successfully hit the infrastructure
to some degree, they are more about people and target, an organization needs strong IT capabili-
processes. It is important to have metadata that ties, a high-level IT champion, and a highly
supports both the IT people who get data in and scalable technical platform. In contrast, hitting
users who get data out. And unless the ware- the transformation target requires top manage-
house contains high quality data, it will not be ment’s business-level support and a BI-oriented
used in the long term. Governance includes the strategic vision.
people, committees and processes that ensure
that BI meets organizational goals. organizational transformation:
Harrah’s entertainment inc.
tHree tarGets for Bi Harrah’s Entertainment Inc. provides a great
example of using BI to support organizational
Organizations make varying levels of commit- transformation. Its CEO, Gary Loveman, a
ment to BI. Some have a few data marts with former Harvard marketing professor, saw the
applications that meet very specific, important opportunity for a new business model when
purposes, such as campaign management, prof- the gaming laws changed. He put together a
itability analysis, and Internet behavior analysis. team of people, including CIO and Director
The other extreme includes organizations that of Strategic Marketing John Boushy, who
have invested in multi-million dollar enterprise shared his vision and created the technology
data warehouses that support strategic business infrastructure needed to support that vision.
objectives. Once the infrastructure was in place, Harrah’s
We believe there are three targets that hired people and developed applications to
organizations can aim for when implementing support fact-based decision making and become
BI (Goodhue, Wixom, & Watson, 2002). Some customer-centric. The organization’s business
companies have a specific need and put specific model and culture changed. The effectiveness
applications in place. For example, a department of the new technology-enabled business strategy
may have a clear business need to run effective is evidenced by Harrah’s current leadership
marketing campaigns, and may implement a position in the gaming industry.
data mart and campaign management software In the 1990s, significant changes took place
and applications to address this need. Other in the gaming industry (Watson & Volonino,
companies aim to create an infrastructure for 2002). New legislation allowed gambling on
BI by cleaning up and defining their data, river boats and Indian reservations, which
establishing efficient processes to move data opened up new markets. Along the Las Vegas
from source systems to a highly extensible data Strip, new properties opened, many with lavish

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 17

Table 1. Three BI targets and their characteristics

Single or a Few Organizational

BI Infrastructure
Applications Transformation
Strategic Vision Satisfy a business unit Provide an organization-wide Fundamentally change how the
need resource business is run
Focus Applications that satisfy Infrastructure that is used Supports and enables a new
particular business units by applications across the strategic business model
needs organization
Level of commitment Low to medium High Very high
Scope Business unit Organization wide Organization wide
Governance Business unit All business units that use the Organization wide, with sig-
infrastructure nificant senior executive in-
volvement
Sponsorship Business unit CIO and business units All C-level executives
Required resources Low to medium High Very high
Impact on people and Limited to people who Makes jobs and processes Fundamentally changes peo-
Processes use the applications more analytical, resulting in ples’ jobs, work processes, and
fact-based decision making the organizational culture
Benefits Low to high at the busi- Provides the infrastructure that Makes the new strategic busi-
ness unit level can generate high returns ness model possible

hotels, shopping malls, and attractions, such rewarded customers for their stay and play at
as a replica of the Eiffel Tower at Paris and the Harrah’s properties.
Bellagio hotel’s dancing waters. To implement this strategy, Harrah’s cre-
Management at Harrah’s recognized that ated a BI infrastructure and applications. Their
there were great opportunities for growth, and starting point was to build a Winners Information
started an ambitious program of building new Network (WINet). WINet sources data from
properties and acquiring existing ones (e.g., the casino, hotel and special event systems.
Rio, and, more recently, Caesar’s). The strategy Customers who join Total Rewards receive a
was to operate the hotels and casinos in an in- loyalty card that they enter into slot machines or
tegrated way and encourage customers to play show to pit bosses when they play table games
at a Harrah’s property whenever and wherever like blackjack. As a result, the casino system
they gambled. This was a significant change captures data on what games people play, for
in Harrah’s business strategy. In the past, Har- what amounts of money, and for how long. The
rah’s hotel and casino managers operated their hotel system captures personal preferences data,
properties as independent fiefdoms. such as whether a customer prefers smoking or
But how could customers be incented to non-smoking rooms. Harrah’s also collects data
seek out Harrah’s casinos? Management ruled about customers who attend special events such
out building lavish hotels, shopping malls, as slot machine tournaments and wine-tasting
and attractions: while taking such actions may weekends. From this data, Harrah’s learns a
draw crowds, they are costly and negatively tremendous amount about its customers.
affect the bottom line. The strategy Harrah’s The data is entered into an operational data
decided upon was to implement BI-based store called the patron database. It is used for a
customer relationship management (CRM) variety of purposes, including generating offers
and an innovative Total Rewards program that to visit a Harrah’s casino (e.g., free gaming chips,

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
18 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

meals, and shows) and providing current data An interesting example of how Harrah’s
to customer-facing applications. For example, was able to capitalize on its investment in BI is
front desk personnel can access a wealth of provided by events that occurred shortly after
information about customers, including their 9/11. Las Vegas is a destination city to which
profitability, room preferences, and recent stays most people travel by plane. Right after 9/11
at a Harrah’s property. Pit bosses can see that the Las Vegas casinos suffered a dramatic drop
a customer is playing far more than usual and in business as people chose not to fly. Quickly
provide a free show ticket. recognizing this, Harrah’s used its patron da-
Some of the data from the patron database tabase and marketing workbench to identify
is loaded into the marketing workbench, which customers who were candidates for driving
serves as Harrah’s enterprise data warehouse. to Las Vegas, and sent them attractive offers.
Analysts use the data in the marketing work- Unlike other casinos, Harrah’s business quickly
bench for their analyses. Many of the applica- returned to normal.
tions help analysts better understand Harrah’s The potential value of a data warehouse
customers. For example, Harrah’s estimates the grows as it adds subject areas. Harrah’s started
value it expects to generate from its customers with customer data and later added product data
over the course of their lifetimes. It identifies (i.e., the games that customers play). With this
their place in a customer lifetime cycle. It additional data, Harrah’s better understands
identifies customer segments for marketing which games various customer segments and
campaigns. Here’s a typical example: A young individuals prefer. For example, it makes sense
male visits a Harrah’s casino for the first time to place slot machines popular with the high
and signs up for the Total Rewards program. rollers close to the high stakes tables. More
The data collected from the casino system re- recently, Harrah’s has been using customer data
veals that he is a high-velocity player, in that he in pricing its hotel rooms as part of a revenue
plays quickly and for relatively high amounts. management initiative that strives to optimize
A further analysis of the data classifies him as the value of its hotel room inventory.
being in the potential growth stage of the cus- Through its innovative use of BI, Harrah’s
tomer life cycle and that he has a high expected has become a leader in the gaming industry.
lifetime value. As a result of these insights, he This formerly modest, blue collar casino is
is sent an exceptionally attractive offer to visit now the largest and most successful gaming
a Harrah’s property soon. company in the world. Its success has caused
Harrah’s is constantly running experiments other companies to copy its most discernable
with control groups to determine what offers BI methods.
are most appealing to different market seg-
ments. Harrah’s Tunica, Mississippi casino, for the Movement to real-time Bi
instance, draws many customers from Jackson.
Which offer is most appealing to these custom- While BI applications may vary from being
ers: one that involves a free night’s stay and a model-based (e.g., revenue management) to
free show, or one that features $60 in free chips? data-based (e.g., dashboards), they all depend
By running an experiment with both offers and on a data infrastructure. This infrastructure has
tracking the results, Harrah’s has learned that evolved over the years, from being application-
the less expensive offer - that of free chips - centric (that is, the data was organized around a
generates the greatest response. It turns out that single or a few applications), to enabling more
people are enticed by the offer of free chips, and traditional data warehousing (that is, the data is
prefer to sleep in their own beds. Prior to the organized as an enterprise-wide resource that
use of BI, decisions about offers would have supports a wide variety of applications), to
been based on intuition, which may or may not today’s real-time data warehousing (Watson,
have been correct. 2005). The recent movement to real-time or

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 19

right-time decision support data represents a Fourteen years ago, senior management
true paradigm shift. implemented a strategy called the Go Forward
Previously, most decision support was Plan, which still drives the company today.
conducted off-line by analysts operating with The four-pronged strategy focuses on reli-
historical data who reported their findings to able operations, a collaborative work culture,
others. With real-time capabilities, an analysis savvy financial management, and value-added
of real-time (and historical data) can drive customer services. The key to this strategy is
operational decisions. For example, decisions information, and Continental has invested in
about which airline passengers should be given an enterprise data warehouse as the platform
a flight upgrade can be based on both customer for delivering key information about customers
value (i.e., profitability) and their recent flight and the business to employees at all levels of
experiences (e.g., a customer who has had flight the organization.
delays might be given an upgrade). Real-time Initially, the data warehouse delivered ac-
data blurs the distinction between operational curate, integrated flight schedule data, customer
and decision support applications. Continental data, inventory data, and other data relevant to
Airlines is a company that is generating great strategic decision-making within revenue man-
value for itself and its customers through real- agement and marketing. Over time, however,
time BI. Continental realized that revenue and marketing
data could be combined with real-time flight
data and used to support important operational
real-tiMe Bi: decisions. For example, airplanes send data in
Continental airlines real-time via satellite to a special interface that
supports the command center for Continental
Continental Airlines has done a remarkable job
where the actual flights are coordinated. This
of adapting to accommodate real-time BI, and
data communicates exactly when a plane pushed
the company is reaping the benefits (Anderson-
off from the gate, how fast the plane is flying,
Lehman, Watson, Wixom, & Hoffer, 2004).
its current location, and when it will arrive at
Real-time applications have found their way
its destination given current conditions. Op-
into operations, marketing, fraud prevention,
erations managers combined plane data with
security, customer service, and other areas. On
marketing data about high value customers
several occasions, its data warehouse has even
and began making decisions about things like
served as a backup for Continental’s transaction
flight delays and gate changes that optimized
systems. Continental continues to be a leader in
customer service interactions.
its use of real-time BI, a capability that likely
Continental discovered that the shift to
contributes to FORTUNE magazine’s recogni-
real-time BI is not straightforward; it requires
tion of Continental as the top airline in its annual
changes to people, process, and technology,
airline industry list of the World’s Most Admired
as well as a new way of supporting BI. For
Companies for the past six years.
example, BI end users changed from a small
Continental is the seventh largest com-
group of highly skilled analytical strategists to
mercial airline in the world, employing more
literally any employee of Continental, from the
than 43,000 people and serving approximately
CEO to tax accountants to flight attendants. The
65 million passengers each year. Continental,
latter, for example, could pull up a BI applica-
along with Continental Express and Continental
tion at the boarding gate to look up high-value
Connection, offers 2,300 daily departures to
customers on a given flight to see if they had
more than 265 destinations across five conti-
any special concerns or preferences. Armed
nents. This year, Continental is celebrating its
with this information, flight attendants could
75th year of business.
then tailor those customers’ flight experiences
appropriately.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
20 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

As the base of end users diversified and when they replace manual reporting processes.
expanded, Continental found itself with a These kinds of benefits can be anticipated,
growing need for employees who understood tracked, and captured in a spreadsheet. Other
BI and had the analytical skills to help increase benefits, such as the enabling of new ways
its use within their areas. We consider IT-savvy of doing business, are much more difficult
business users hybrid employees – business to quantify, but may generate a competitive
users who take on traditional IT roles such as advantage or open up new markets for the
training, report development, and requirements company. When BI generates these kinds of
determination. The spread of BI to operational benefits, companies spend less time justifying
decisions also prompted process changes as their annual BI investments, focusing instead
BI tools became part of everyday business on crafting new and creative ways to leverage
processes. Tax accountants, for example, use and evolve their BI capabilities.
queries to replace manual checks and balances, Figure 2 illustrates the wide range of pos-
resulting in faster, more accurate tax manage- sible benefits resulting from BI. Note that the
ment. In the United Kingdom, Continental most tangible and easy-to-measure benefits have
must pay a departure tax for passengers who more of a local impact, typically happening at
leave on Continental flights. Manual processes the departmental level. The more intangible
could not always identify exceptions to the tax benefits – things like process improvement and
rules. Using BI, the tax accountants were able strategic enablement – can have impacts across
to eliminate overpayment of the departure tax, an organization. Throughout our case studies,
resulting in annual savings of $300,000 for we have observed that there is an association
Continental. between the kinds of benefits realized by BI,
Other significant changes from real-time and an organization’s BI target. As Table 2
BI occurred within the IT department. As BI shows, companies often get what they pay for.
applications became more and more mission Localized benefits are associated with localized
critical, and more deeply embedded into ev- initiatives, which are much less costly to put
eryday processes, the data warehouse and its in place than enterprise BI programs that result
applications began to require treatment more in significant, transformational impacts. In the
like that of transaction systems. IT needed to following section, we describe the portfolio of
provide 24/7 support, plan for disaster recovery, benefits being realized at Norfolk Southern.
and ensure fast (i.e., operational) performance.
To do this, IT staff needed operations experience Bi Benefits at norfolk southern
combined with knowledge of BI. In addition,
the structure of the IT organization (e.g., which Norfolk Southern is one of the four largest
group should be in charge of the customer data freight railroads in the United States. Each
hub) suddenly became far more complicated. day, the company moves about 500 freight
trains across 21,000 route miles in 22 eastern
states, the District of Columbia, and Ontario,
Benefits of Bi Canada. Norfolk Southern manages more than
$26 billion in assets and employs over 30,000
While BI may generate new complexities, it people.
also, of course, generates a wide variety of Today, Norfolk Southern is home to a
organizational benefits (Haley, Watson, & six terabyte Teradata data warehouse, which
Goodhue, 1999). Some BI benefits are tangible manages an extensive amount of information
and easy to measure. For example, companies about the company’s vast network of railroads
may eliminate software and hardware licenses and shipping services. Norfolk Southern uses
and fees when they consolidate and retire data the data warehouse to analyze trends, develop
marts, or companies may reduce headcount forecasts and schedules, archive records, and

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 21

Figure 2. Benefits of BI

Table 2. Three BI targets and their costs and benefits

Single or a Few Organizational

BI Infrastructure
Applications Transformation
Costs Costs are relatively low. Infrastructure creation is An expensive, risky un-
costly, requires a vision to dertaking.
justify. Departments give up
control of their data.
Benefits Greater efficiencies and Possible cost savings from The potential for great
effectiveness at the depart- infrastructure consolida- increases in revenues and
mental level may result in tion. Possible quick-hit profits.
greater local revenues and returns from follow-on
profits. applications.

facilitate customer self-service. The benefits of sets. Then, in 1980, the industry was partially
the warehouse range from the highly tangible— deregulated, which opened up opportunities
i.e., cost savings—to the far less tangible—i.e., for mergers and allowed companies to charge
improvements in decision-making and support rates based on service and enter into contracts
for strategic goals. Notably, departments across with customers. Delivering shipments on time
the enterprise routinely realize these benefits. became an important factor in the industry.
But this sort of BI-based efficiency is a rela- Norfolk Southern recognized that a
tively new development for Norfolk Southern. competitive shift was in order, and decided to
For more than a century, the railroad industry become a scheduled railroad. This meant that
was heavily regulated, and Norfolk Southern Norfolk Southern would develop a fixed set of
made money by managing its costs. Managers train schedules and a fixed set of connections
focused on optimizing the use of railcars to for cars to take as they move among trains and
get the most production out of their fixed as- yards. In this way, managers could predict when

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
22 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

they could get a shipment to a customer. The to operations would actually strategically repo-
ultimate intent was to compete in the market- sition the company. In 2002, Norfolk Southern
place based on service. expanded the use of its data warehouse to sup-
At the time, Norfolk Southern improved port an application for its railroad operators.
its service to customers by providing them The Thoroughbred Operating Plan (or TOP)
with information about their shipments. Prior dashboard application pulled data from the data
to 2000, customers would call or fax a Norfolk warehouse and then graphically depicted actual
Southern customer service agent with questions, performance against the trip plan for both train
and then wait for minutes, hours, or days for an performance and connection performance. The
answer. Behind the scenes, agents had to place application used visualization technology so that
information requests into the IT department, or field managers could more easily interpret the
navigate legacy systems that were hard to use. large volumes of data (e.g., there were 160,000
The customer service processes were costly weekly connections across the network). The
and time-consuming. number of missed connections has fallen sixty
Norfolk Southern implemented a one percent since this application was implemented.
terabyte data warehouse to store data about And in the past five years, rail car cycle time
railcars, shipments, human resources, and has decreased by an entire day, resulting in
other key transactions. In 2000, the Marketing millions of dollars in annual savings.
department began to explore using the Internet Norfolk Southern has an enterprise data
to phase out telephone- and fax- based informa- warehouse, which means that once data is
tion services, and found that the data warehouse placed in the warehouse, it is available across
could serve as a platform for serving customers. the company, not just for a single application.
They built a BI application called accessNS Although train and connection performance data
that allowed customers to inquire about their is used for AccessNS and the TOP application,
shipments using a Web interface. Customers the company has been able to leverage that data
could determine where their shipments were for all kinds of other purposes. One interesting
right now – and answer questions about their internal application was developed by Norfolk
shipment’s history: Where did my shipment Southern’s human resources department to
come from? How long did it take to arrive? help them make better decisions. Recently,
What were the problems along the route? This the department needed to determine where to
departmentally focused application had clear, locate service offices in the field to best meet
tangible benefits: time savings and improved the needs of the company’s 30,000+ employees.
customer satisfaction resulting from more and By combining employee demographic data (e.g.,
better information. zip codes) with geospatial data traditionally used
Today, accessNS allows more than 14,500 by the engineering group, human resources was
users from 8,000 customer organizations to log able to visually map out the employee popula-
in and access predefined and custom reports tion density, making it much easier to optimize
about their accounts at any time. The users can service office locations.
access current data, which is updated hourly, or
they can look at three years of history. accessNS Bi Governance
provides alerting and RSS feed capabilities; in
fact, 4,500 reports are delivered to users daily. Not surprisingly, governance is a critical compo-
Norfolk Southern has calculated that it would nent of any BI initiative. The more BI-based an
require 47 people to send out today’s volume of organization is, the more comprehensive its BI
reports using the old manual processes. governance should be. For rich BI environments,
Norfolk Southern recognized that transpar- the scope of governance should vary from the
ency into operations would improve customer strategic to the operational. At its highest level,
satisfaction, but it turned out that improvements governance ensures that the BI and business

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 23

strategies are aligned. It should prioritize proj- the BI staff chairs and facilitates the activities
ects and make the required resources available. of each of the various groups.
At lower levels, it should ensure that there are The highest-level committee is composed
consistent data definitions. of business leaders from each of the major
There is no single model for BI governance. divisions within the organization. They meet
What works best depends on the organization’s as needed and provide high-level direction for
BI target, which projects warrant special atten- BI initiatives. For instance, they offer strategic
tion at a given point in time, the organization’s guidance, perform high-level resource manage-
overall approach to IT and business governance, ment, and help resolve project prioritization
and other factors. Let us consider the best prac- issues. This group focuses on ensuring the
tice approach to BI governance used at Blue alignment of BI with corporate direction.
Cross and Blue Shield of North Carolina. Until recently, the mid-level committee
was composed of directors and managers who
prioritized projects and allocated resources. It
Bi GoVernanCe at Blue also resolved cross-functional area conflicts.
Cross and Blue sHield This group has changed recently, as BCBSNC
of nortH Carolina is rolling out a next-generation data warehouse.
The warehouse will be more comprehensive
Blue Cross and Blue Shield of North Caro-
and easier to use. It will include mechanisms
lina (BCBSNC) delivers health care products,
that will allow customers to access data to
services, and information to its more than 3.7
help them better manage their own healthcare.
million North Carolina members. It provides an
It will have multidimensional dependent data
excellent example of effective BI governance
marts to make it easier for users to understand
(Watson, Fuller, & Ariyachandra, 2004). The
and access data. There are also initiatives to
governance committees are multi-level and
improve metadata and data quality. Because
cross-functional, and the right people are present
the warehouse is in building mode with a large
to discuss the important issues. The approach
number of projects to support the initiative,
to governance is dynamic; it is adjusted to best
the mid-level committee has been disbanded
meet current needs, such as restructuring the
and replaced with cross-functional teams that
mid-level committee to more of a project man-
work on specific projects. Once the projects are
agement focus to support the next-generation
completed, BCBSNC may reinstate the original
warehouse.
mid-level committee.
BCBSNC has been a BI-based organization
The lowest-level committee is composed
for over ten years. The company has stated that
of power users and representatives from the
its BI mission is to be “the engine that pow-
business units that use BI. This group meets
ers a customer-focused, information-driven
bi-weekly to discuss and communicate BI de-
company.” BCBSNC has an enterprise data
velopment and use issues with the BI staff and
warehouse that sources data from over a dozen
other users. Much of the focus is on the data,
systems and supports applications in Market-
including such issues as data quality.
ing, Financial Services, Corporate Analysis and
Risk Assessment, Corporate Audit, and Health
Quality Improvement. a Bi MaturitY Model
A multi-level governance structure has
played a key role in ensuring that BI is meeting Maturity models help define and categorize the
business needs and adapting to organizational state of an organizational capability. Maturity
change. To ensure effective communication models have a long history in the IS field (Wat-
among the different committees, a member of son, Ariyachandra, & Matyska, 2001). The first
and most famous maturity model was Gibson

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
24 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

and Nolan’s (1974) four stages of EDP growth, perceptions, types of analytics, stewardship,
which describes the evolution of planning, funding, technology platforms, and change
organizing, and controlling an organization’s management and administration. Eckerson’s
computer resources. Maturity models also are model provides a useful lens for understanding
sometimes called stage or phase models, and BI-based organizations.
they have been especially popular with prac- We have discussed three specific targets
titioners because of their perceived usefulness for BI – a single or a few applications, BI
and face validity. infrastructure, and organizational transforma-
The fundamental concept underlying tion. Each of these targets can be mapped onto
maturity models is that things change over Eckerson’s maturity model. Companies that
time, but in sequential, predictable ways. The develop a single or a few applications are at
various stages are defined and identified by a an early stage of BI evolution, typically at the
set of characteristics, which have their own infant or child stage. The characteristics of this
maturity cycles. For example, funding may be stage—the vision for BI, funding, platform and
a component characteristic of a larger maturity data management, and governance—all tend
model and have its own stages. to be at the departmental or work unit level.
Though companies move through stages There are benefits to be realized, but they tend
in sequential, predictable ways, the movement to be localized.
is not always easy. To move from one stage to The BI infrastructure target maps onto the
another requires changes in all of the charac- teenager and adult stages. There is sponsorship
teristics that make up the stages. This typically at the CIO and senior management levels. BI
includes changes in management vision, fund- and data are viewed and treated as a strategic
ing, data management, and more. While all of resource. One or more data warehouses provide
the stages for the characteristics do not have to a single version of the truth. There is a rich and
be exactly in synch, they should be at approxi- varied set of analytical applications that help
mately the same stage of evolution. drive the business. Enterprise-wide BI gover-
Companies sometimes try to skip stages, nance is in place. From an ROI perspective,
but this seldom proves to be wise. There is the investment in BI really pays off. Currently,
valuable experience to be gained at each stage, many firms are at this point in their BI evolution.
and attempting to move too quickly through A 2007 study by TDWI found that 64 percent
the multiple stages of evolution is not only of the respondents were at the teenager stage
difficult for an organization, but also can have (Eckerson, 2007).
negative repercussions. A notable exception is The organizational transformation target
when an organization is facing a major crisis, or is associated with the adult and sage stages.
when its survival is at stake. In such situations, BI is well-established, organization-wide, and
management is often willing to do whatever it mission critical. As we describe it, however,
takes – funding, outside help, reorganizations – the decision to make organizational transforma-
to move to an advanced stage of maturity. tion a target involves consideration that goes
There are a number of maturity models for beyond the characteristics that typically define
BI. A popular one was developed by Wayne the stages. Management recognizes that BI is a
Eckerson (2004), Director of Research for The strategic enabler necessary to take advantage of
Data Warehousing Institute, the leading profes- a once-in-a-lifetime market opportunity, or to
sional organization for BI practitioners. Ecker- respond to an organizational crisis. The former
son’s model uses a human evolution metaphor was the case at Harrah’s Entertainment. Though
to describe six stages: prenatal, infant, child, not discussed here, the latter is illustrated at First
teenage, adult, and sage. The stages are charac- American Corporation, a bank that ensured its
terized and defined by a set of characteristics, survival by implementing a BI-based customer
including scope, analytic structure, executive intimacy strategy (Cooper, Watson, Wixom,

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 25

& Goodhue, 2000). While BI can help to cre- simply not applicable. At Continental Airlines,
ate significant financial benefits, its creative for example, North American revenue manage-
potential extends far beyond the realm of the ment models are not used in Japan because
financial. BI lies at the core of firms’ successful Japanese ticket sales through travel agents have
execution of strategy, and how they compete unique behavior patterns, and the mainstream
in the marketplace. analytics are not helpful.
Multi-national companies and vendors
that sell BI products and services globally need
ConClusion to understand what practices can and cannot
be replicated around the world, and for what
Although much is known about BI, and con-
reasons. Ultimately, academics need to explore
ceptual foundations and frameworks (including
global BI with the objective of discovering how
the ones presented in this article) exist, the BI
to better diffuse and tailor BI across diverse
field offers a wealth of research opportunities
global settings.
for academics to explore. And, new avenues
for research regularly arise. We will close by Bi for decision Making
describing a sampling of underexplored and
unexplored research topics that exist for BI BI is implemented to help decision makers
academics today. We hope that over time, the make better decisions, but there is still much to
International Journal of Business Intelligence learn about the design of BI for decision mak-
Research will showcase work in these areas. ing, where BI fits within the decision making
process, and, ultimately how and why BI makes
Global Bi a difference.
A critical facet of BI design is the BI
The use of BI is uneven around the world
interface, which includes the display of data.
(Watson and Swift, 2002). In North America
Edward Tufte (2001) suggests that the display of
and Northern Europe (i.e., Norway, Sweden,
information significantly can impact the way in
and Finland), BI is well entrenched, as it is
which data is interpreted and applied. Although
in Hong Kong. This is less true in Central
some studies empirically have compared the
and South America and especially in Africa.
impact of various interface designs within a
These differences may result from dissimilar
context of decision support (e.g., Dos Santos
states of economic development, competitive
& Bariff, 1988; Todd & Benbasat, 1992), much
forces, or cultures, among other reasons. The
more work is needed. As new displays of data
global diffusion of BI needs to be investigated
become popularized in BI, such as heat maps
empirically.
and animation, we need to understand how to
For example, customer relationship
leverage them appropriately.
management is a popular BI application that
At Norfolk Southern, system developers
requires the collection and use of transaction,
spend a great deal of time researching and
demographic, financial, and psychographic data
testing out the designs of BI user interfaces.
to drive customer interactions. A major event
They believe that the BI interface impacts user
such as the purchase of a home or retirement
acceptance and use, as well as decision making
may trigger a set of offers that are tailored to
outcomes; however, it is still unclear exactly
that person. In some counties, such as China,
what designs best support various BI needs. As
there are culturally determined hesitancies
BI becomes more pervasive and companies at-
about collecting and using personal data; some
tempt to deliver BI applications to the masses,
countries are bound by legal constraints (e.g.,
including customers and suppliers, there is a
privacy laws); and some countries find that
greater need to understand how to design BI to
behavioral norms from other geographies are
be easy to use and to be applied correctly.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
26 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

Also, practitioners need a better under- centers (Zeid, 2006) effective ways for building
standing of how metadata should be incorpo- organizational capabilities in BI?
rated into interfaces and the BI infrastructure. Further, as BI becomes more pervasive
Practice recognizes that metadata is critical for within the enterprise, practitioners need to
BI because system users and designers need to understand how to roll-out BI to larger audi-
understand the meaning and nuances of data ences in cost effective ways. Will investments
before it can be used in appropriate ways (Fo- in training, communication strategies, interface
shay, 2007). More academic work addressing designs, or process change help pervasive BI
metadata can help further our understanding efforts deliver greater value to the organization?
of what metadata strategies to apply for BI How can BI be embedded within front-line
and why. processes as well as emerging social network-
Much is known about the decision making ing communities?
process in general and important concerns that These few areas – global BI, BI for decision
adversely impact the decision making process making, and organizing for BI – are among many
(e.g., managerial biases, information overload) areas that are ripe for exploration by academics
(Bazerman, 1986; Hammond, Keeney, & Raiffa, interested in furthering our understanding of the
2006). Given this understanding, academics field. It is fascinating to consider the progress
need to help translate existing knowledge about made on the path from decision support in the
decision making into structures, processes, 1960’s to BI today. But, even more fascinating
and designs that improve the effectiveness of is the opportunity for academics who seek to
BI. And, as BI data moves from being highly further our understanding of the future.
quantitative to encompassing documents, video,
and other unstructured data formats, research
is needed to understand how to use this data to referenCes
make decisions – and whether emerging tech-
nologies like BI search engines, visualization, Anderson-Lehman, R., Watson, H. J., Wixom, B. H.,
text mining, mash-ups, and semantic webs help & Hoffer, J. A. (2004). Continental Airlines Flies
High with Real-Time Business Intelligence. MIS
improve decision outcomes. Quarterly Executive, 3(4), 163–176.

organizing for Bi Ariyachandra, T., & Watson, H. (2005). Key factors

in selecting a data warehouse architecture. Business
Traditionally, IT organizations contain a group Intelligence Journal, 11(1), 4–6.
(or groups) of employees dedicated to delivering Bazerman, M. H. (1986). Judgment in Managerial
decision support applications and maintaining Decision Making. New York: Wiley.
data warehouse infrastructures. There are sev-
Cooper, B. L., Watson, H. J., Wixom, B. H., & Good-
eral trends that challenge separating BI into its hue, D. L. (2000). Data Warehousing Supports Cor-
own organization structure(s). For example, the porate Strategy at First American Corporation. MIS
movement to real-time BI transforms decision Quarterly, 24(4), 547–567. doi:10.2307/3250947
support into more of an operational function, Davenport, T. (2006, January). Competing on Analyt-
which requires operational service levels, skill- ics. Harvard Business Review, 2–10.
sets, and structures. IT leaders are struggling
with how to organize their IT organizations to ad- Dos Santos, B. L., & Bariff, M. L. (1988). A Study
of User Interface Aids for Model-oriented Decision
dress this shift. Research is needed to understand Support Systems. Management Science, 34(4),
how changing IT roles and capabilities translate 461–468. doi:10.1287/mnsc.34.4.461
into new kinds of organizational designs and
Eckerson, W. (2002). The Decision Support Sweet
processes. Should areas of BI (e.g., the data
Spot. Journal of Data Warehousing, 8(3), 4–9.
warehouse group) be folded into operational
units within IT? Are designs like BI competency

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010 27

Eckerson, W. (2004, November). Gauge Your Data Tufte, E. R. (2001). The Visual Display of Quantita-
Warehouse Maturity. DM Review, 51, 34–37. tive Information (2nd ed.). Cheshire, Connecticut:
Graphics Press.
Eckerson, W. (2007). 2007 TDWI Benchmark Re-
port. Seattle, Washington: The Data Warehousing Watson, H. J. (2005). Real Time: The Next Generation
Institute. of Decision-Support Data Management. Business
Intelligence Journal, 10(3), 4–6.
Foshay, N. (2007). Does Data Warehouse End-User
Metadata Add Value? Communications of the ACM, Watson, H. J., Ariyachandra, T., & Matyska, R. J. Jr.
50(11), 70–77. doi:10.1145/1297797.1297800 (2001). Data Warehousing Stages of Growth. Infor-
mation Systems Management, 18(3), 42–50. doi:10.
Gartner (2008). Gartner EXP Worldwide Survey of 1201/1078/43196.18.3.20010601/31289.6
1,500 CIOs Shows 85 Percent of CIOs Expect Sig-
nificant Change Over Next Three Years. Retrieved Watson, H. J., Fuller, C., & Ariyachandra, T. (2004).
May 20, 2009, from http://www.gartner.com/it/page. Data Warehouse Governance: Best Practices at Blue
jsp?id=587309 Cross and Blue Shield of North Carolina. Decision
Support Systems, 38(3), 435–450. doi:10.1016/j.
Gartner (2009a). Gartner EXP Worldwide Survey dss.2003.06.001
of More than 1,500 CIOs Shows IT Spending to Be
Flat in 2009. Retrieved May 20, 2009, from http:// Watson, H. J., & Swift, R. (2002). Data Warehousing
www.gartner.com/it/page.jsp?id=855612 Around the World. Journal of Global Information
Technology Management, 5(2), 1–6.
Gartner (2009b). Howard Dresner. Retrieved May
20, 2009, from http://www.gartner.com/research/ Watson, H. J., & Volonino, L. (2002). Customer
fellows/asset_79427_1175.jsp Relationship Management at Harrah’s Entertain-
ment. In G.A. Forgionne, J.N.D. Gupta, & M. Mora
Gibson, C. F., & Nolan, R. L. (1974). Managing (Eds.), Decision-Making Support Systems: Achieve-
the Four Stages of EDP Growth. Harvard Business ments and Challenges for the Decade (pp. 157-172).
Review, 52(1), 28–40. Hershey, PA: Idea Group Publishing.
Goodhue, D. L., Wixom, B. H., & Watson, H. J. Williams, S. (2004). Assessing BI Readiness: A
(2002). Realizing Business Benefits through CRM: Key to BI ROI. Business Intelligence Journal,
Hitting the Right Target in the Right Way. MIS 9(3), 15–23.
Quarterly Executive, 1(2), 79–94.
Zeid, A. (2006). Your BI Competency Center: A
Haley, B. J., Watson, H. J., & Goodhue, D. L. Blueprint for Successful Deployment. Business
(1999). The benefits of data warehousing at Whirl- Intelligence Journal, 11(3), 14–20.
pool. Annals of Cases on Information Technology
Applications and Management in Organizations,
1, 14–25.
Hammond, J. S., Keeney, R. L., & Raiffa, H. (2006). endnote
The Hidden Traps in Decision Making. Harvard
Business Review, 84(1), 118–126. 1
It should be pointed out that some authors use
the term BI to refer to getting data out, and the
Power, D. J. (2007). A Brief History of Decision term data warehousing to refer to getting data
Support Systems. DSSResources.COM Version 4. in. There are also some authors who use the
Retrieved May 20, 2009, from http://DSSResources. term data warehousing to refer to both getting
COM/history/dsshistory.html data in and getting data out, much like we are
Sprague, R. H. Jr. (1980). A Framework for the using the BI term. The good news is that the
Development of Decision Support Systems. MIS differing terminology does not normally cause
Quarterly, 4(4), 1–26. doi:10.2307/248957 confusion because of the context in which the
terms are used.
Todd, P., & Benbasat, I. (1991). An Experimental
Investigation of the Impact of Computer Based Deci-
sion Aids on Decision Making Strategies. Informa-
tion Systems Research, 2(2), 87–115. doi:10.1287/
isre.2.2.87

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.
28 International Journal of Business Intelligence Research, 1(1), 13-28, January-March 2010

Barbara H. Wixom is an associate professor in the McIntire School of Commerce at the University
of Virginia. She is a fellow of The Data Warehousing Institute and an associate editor of the Journal
of BI. Her research and teaching focuses on business intelligence success and benefits.

Hugh J. Watson is a professor and holder of a C. Herman and Mary Virginia Terry Chair of
Business Administration in the Terry College of Business at the University of Georgia. He is the
author of 22 books and over 100 academic journal articles. Watson is a fellow of the Association
for Information Systems and The Data Warehousing Institute.

Copyright © 2010, IGI Global. Copying or distributing in print or electronic forms without written permission of IGI Global
is prohibited.